COST OF MODEL-REFERENCE ADAPTIVE-CONTROL - ANALYSIS, EXPERIMENTS, ANDOPTIMIZATION

In this paper a study of the performance of model reference adaptive c ontrol is presented in numerical simulations and verified experimental ly on a single-degree-of-freedom system with the objective of understa nding how differences between the plant and the reference model affect the maximum control force and control effort performance metrics. The se performance metrics were found to be highly sensitive to difference s between the plant and the reference model. It was also found that wh en such differences exist, lowering the required performance by reduci ng the required damping actually increases the maximum force and contr ol effort for a single-degree-of-freedom example. Optimization of weig hting matrices is shown to help reduce the increases in these performa nce metrics, However, the optimization results in a design that, for s uccessful realization, requires a sampling frequency 40-80 times the N yquist frequency,